290 Life and Letters of Francis Galton
of the axial four positions of this card and found : (i) a sharply defined cross, showing the accuracy of adjustment, (ii) four very faint dots one in each quadrant, indicating that any single irregularity hardly survives, (iii) the equal tint of the four dots showing the equality of the exposure, (iv) the uniformity of the resulting tint of the four `wafers' arising from the exposure of the tinted discs, A, B, C, D, in the four orders, ADCB, BADC, CBAD, and DCBA, demonstrating that order of exposure is not material.
He also showed composites in one of which a portrait X was exposed g and a portrait Y . of the total exposure, and in the other X was exposed
and Y I of the total. The individuality of X predominated in the first and that of Y in the second ; thus justifying `weighting' by' length of exposure.
He further exhibited composite portraits of male and female phthisical subjects, and of men and of officers of the Royal Engineers (see our pp. 291-3 and Plates XXIX and XXXIV), and he suggested that with 'artistic touching' beautifully idealised family portraits might be produced for commercial purposes; the irregularities of the individual disappearing. The paper as a whole marks a very considerable development both in the theory and practice of composite portraiture'.
We now come to two papers of 1882 and 1885 dealing respectively with composite photographs of phthisical subjects and of Jews. In the first Galton worked in conjunction with Dr F. A. Mahomed, in the second Mr Joseph
1 It was, perhaps, a misfortune for composite photography, that while it required really extraordinary care and patience, it was very easy to compound in an inferior manner. It became popular, especially in America, and a good deal was published which is of small scientific value and in which no attempt was made at real analysis of the results. Thus Science published May 8, 1885, composite portraits of American (a) Mathematicians, (b) Naturalists, (c) Academicians and (d) Field Geologists, which lead us hardly further than the conclusions that all American scientists of those days were hairy, and that mathematicians while being least so had more frown. Composites of Washington. in three aspects (Science, Dec. 11, 1885) are somewhat more successful : Science also published (on May 7, 1886) composites of American Indians, but the components were few in number. Further composite photographs were made of undergraduates and graduates of various American Colleges (Jastrow, 1887, did 21 Johns Hopkins doctors of philosophy for the years 1886-7: Century, March, 1887. The latter also contains a fairly successful family composite of father, mother, five sons and one daughter). A possibly more scientific use of composite photography was that by Persifor Frazer (American Philosophical Society, Vol. xxiii, pp. 433-41, 1888 and Franklin Institute Journal, Feb. 1886, p. 123) to obtain an average signature. He illustrates it by one of George Washington, and thinks the process could be used to determine the maximum deviation compatible with a nonforged result. In our own country Arthur Thomson in 1884 (Journal of Anatomy, Vol. xix, p. 109) tried to apply it to obtain type Australian and European crania; the components being too few, and the superposition not very satisfactory, the results are not to be taken as condemning the application of the method in craniometry. The possibilities of composite photography in this matter had been referred to by Galton at the 1881 York meeting of the British Association: see Transactions, p. 690. Whipple adopted the process for the reduction of meteorological observations (Quarterly Journal of Meteorological Society, Vol. ix, p. 189), and it can clearly be employed in harmonic analysis. There is no doubt that Galton's idea was taken up by many, but it may be doubted whether any one but the originator appreciated the amount of care and patience required to produce a good composite. At the same time I cannot pass over the fact that in the Galtoniana there exists a splendid series (not by Galton) of racial composites, Wends 'and Saxons; they are probably of German origin. I am unaware if they have been published.